What our own test actually measured
We built the test the boring way. Gaps were defined mechanically from closed candles only, entries were placed on a touch of the gap with a fixed stop beyond the originating candle, targets were set at a fixed multiple, and nothing was adjusted after the fact.
Across two years of gold data the result was flat. Winners and losers roughly balanced after costs, and the outcome was not stable across sessions or timeframes, which is the pattern you expect from noise rather than from an edge that happens to be small.
No measurable edge does not mean the gap is not real. It means the mechanical entry on its own does not pay for the spread, the slippage and the stops you collect on the way. Those are different claims and the distinction matters.
What a gap on the gold chart records
An imbalance appears when price moves far enough in one direction that a candle leaves no overlap with the body range two candles back. Nobody transacted across that span at leisure, because the move went through it in one push.
On gold this happens most often in three places: the first minutes after a scheduled release, the London open sweep, and the thin hour after the New York close when a modest order can shift the quote several dollars.
That origin story is the useful part. A gap carved by a payroll release is a reaction to information. A gap carved at midnight in an empty book is a liquidity artefact. Both look identical as rectangles, and treating them the same is the first mistake most traders make with the concept.
Why fills are common and still useless
Gold spends most of its hours rotating inside a range, and intraday rotation fills nearly every nearby void sooner or later. A claim that gaps get filled is therefore close to a claim that price revisits recent prices, which is true of almost any level you draw.
The trading question is narrower and harsher. Does price reach the gap before reaching your stop, does it turn from the gap rather than through it, and does the resulting move cover costs? Those three conditions are where the statistics collapse.
A fill rate quoted without a stop distance, a time limit and a cost assumption is not a result. It is a description of how often two prices met, which the market provides for anything you care to mark.
When gold closes a gap quickly
Fast closure has a profile. It shows up when the gap was created by thin liquidity rather than by new information, when the gap sits inside an established range rather than beyond its edge, and when price is still in the session that created it.
- Gaps formed during the quiet late New York hours often close during the Asian session.
- Gaps inside the prior day range tend to be revisited within a day or two.
- Gaps created by a headline that is later contradicted close almost immediately.
None of these are entries by themselves. They are a reason to expect rotation rather than continuation, which changes how you manage an existing position far more than it justifies a new one.
When a gap stays open for sessions
The opposite case is a gap carved by genuine displacement: a decisive break away from a level with sustained volume behind it and follow through in the next several candles. Those voids can sit untouched while price trends away, and traders who sit waiting for the retrace simply miss the move.
This is why using an unfilled gap as a reason to fade a trend is expensive on gold. During a strong directional phase, driven by a shift in rate expectations, the metal leaves a ladder of unfilled imbalances behind it for weeks.
Judge displacement first and the gap second. If the candle that made the gap also broke structure and held, the gap is a footprint of strength, not a magnet you should be trading against.
Using a gap as context beside a real level
The one use that survives honest testing is confluence. A gap that happens to sit inside a daily support or resistance zone you had already marked adds detail to a decision you were going to make anyway: it narrows where inside the zone you want to act and where the idea is wrong.
That is how we treat it in our tools. The Today Levels zones come from our own level method with a confluence score, and an imbalance that lands inside one is a refinement of location rather than a separate signal. MEER FVG Pro applies a strict rule that the fifteen minute and five minute readings must agree before anything is shown, which exists to slow traders down, not to promise a result.
If you want the full mechanics of drawing and grading these zones, our guide to fair value gaps on gold covers the construction in detail.
What the test could not settle
A flat result from one rule set does not prove every gap method fails. Our test used fixed stops, fixed targets and mechanical entries, and it is possible that a discretionary trader reading context around the same gaps performs better. It is equally possible that such a trader is adding the context and the gap is contributing nothing.
The test also cannot speak to very short holding times inside a single minute, where execution quality and spread dominate everything else and retail results are unlikely to resemble the model.
What it does establish is the default. Absent your own evidence, treat gap entries on gold as unproven and require something else to carry the trade.
How to use this on the chart
Mark imbalances, but label their origin. Write news, sweep or thin hours beside each one, because that single word predicts behaviour better than the rectangle does.
Only act where a gap overlaps a level that existed before the gap formed. Place the stop beyond the structure that created the move rather than at the far edge of the rectangle, since the edge is an arbitrary line and the structure is where the idea is actually invalidated.
Then record what happened. After thirty observations of your own, on your own feed, you will know whether gaps add anything to your decisions on gold, and you will not need to take our result or anybody else on trust. You can mark them live on the XAUUSD chart while you build that record.
FAQ
Do fair value gaps work on gold?
As a standalone entry rule, our two year test on gold data found no measurable edge after costs. Gaps still mark where price moved in one direction without opposition, which is real information about location. Used as a refinement inside a support or resistance zone you had already identified, they help you choose a price. Used alone as a trigger, they did not pay.
Why does gold fill gaps so often then?
Because gold spends most of its time rotating, and rotation revisits almost any nearby price. A high fill rate is close to saying that price returns to where it recently traded, which is true of most levels you could draw. The useful question is whether price reaches the gap before your stop and turns there, and that is where the numbers stop being impressive.
Which timeframe gap is most reliable on gold?
Higher timeframe gaps survive longer because they need a larger move to be closed, so four hour and daily imbalances tend to stay relevant for days. That is durability rather than reliability. Our testing did not find one timeframe that produced an edge, and the lower timeframe gaps suffer most from spread and slippage on gold.
Should I ignore imbalances completely?
No, but demote them. Keep them as a map of where recent moves were one sided, and note what created each one, since a gap from a data release behaves differently from a gap carved in thin overnight liquidity. Let your levels and structure decide whether to trade, and let the gap decide where inside the zone you act.
ⓘ See these ideas on real price: open the free XAUUSD live chart.